October 2024—Do clinicians understand how the technology in hematology has evolved and how laboratory data can help guide their decisions? It’s a question roundtable participants took on when they met online Aug. 29 with CAP TODAY publisher Bob McGonnagle. “There’s a disconnect with our clinical colleagues,” said Olga Pozdnyakova, MD, PhD, of the Hospital of the University of Pennsylvania. She and others spoke about solutions, instruments, AI, and reference ranges—in addition to the staffing shortage. “It is first and foremost in our minds,” Maria (Ria) Vergara-Lluri, MD, of Keck School of Medicine of USC, said of the ongoing shortage.
Following their conversation below is an all-new CAP TODAY guide to hematology analyzers. The questions we asked vendors about their instruments were updated and revised for the first time, providing new insight into the instruments on the market.
A few issues from our discussion in last year’s roundtable on hematology analyzers carry over to today’s conversation: the efforts to cut back on the number of manual differentials we’re asking to have done in labs; the continuing work around bands versus neutrophils; the remote viewing of cells, à la surgical pathology, when we want an automated look at blood cells; and artificial intelligence in the world of hematology. Fernando Chaves, is there something new here? Where do you place us in the state of hematology today?
Fernando Chaves, MD, director, medical and scientific affairs, specialty lab solutions, Siemens Healthineers: Our main focus continues to be on digitalization and digital integration to improve patient care and workflow. For manual reviews, it’s not a question of how many slides are needed but how much labor and turnaround time are required to review a slide. That’s why we focus on the importance of digital morphology, the ability to use digital hematology, which gives a full-field view on a computer screen and the same experience as you have on a microscope. You see not only random selected cells but also a view of the entire slide. You can see the context of each cell, navigate through the slide on the computer screen, and do it remotely. It saves time and shortens turnaround time.
There’s a shortage of qualified staff, an even more acute shortage of expert staff as people retire, and a shortage of pathologists, with hematopathologists probably in the greatest demand by people who are hiring. Sarah Borah, I know you pride yourself at Beckman on your ability to automate and engineer to make things more efficient and take the burden off customers. Tell us what Beckman is working on that you’re excited about.
Sarah Borah, MSBA, MT(ASCP), marketing director for North America, hematology/urinalysis solutions and clinical market development, Beckman Coulter: We’re seeing more and more customers needing to move to automated solutions—automating the steps with our robotics and automation lines—but we’re also incorporating AI into that, which adds more advantages. As senior technologists retire, especially in hematology, being able to spread that expertise across the lab by having remote capability is impactful. We can now take a slide and essentially Google-scan it and customers can look at it remotely. A senior technologist from one location can remote in and help a more junior technologist in another location.
At Beckman Coulter, we want to advance our hematology technology where we can identify and classify cells with AI on the CBC instrument itself. That would give the CBC instrument the ability to help with staff and expertise gaps in the lab. We’re thinking about if there could be a slide-less future in hematology where we don’t have to rely on slide review, a future in which we’re able to get images and classification from the analyzer, removing the need to review a glass slide for most CBC samples.
Kevin Stuteville from Abbott, what are your thoughts as we raise these issues in hematology analysis?
Kevin B. Stuteville, MBA, MT(ASCP), global product manager, hematology, Abbott: At Abbott we have gone full steam ahead with total laboratory automation and integration with robotics, and hematology workflows are part of that now. Hematology is a different fit into the automation and we’re integrating it into our GLP systems Track. I agree with Sarah on the integration of digital morphology into the hematology instrument itself. It’s a fully capable instrument that will do everything for the technologist. Abbott’s Digital Health Solutions branch is looking at integrating AI features into traditional middleware, automation, et cetera.
Susan Behnke, everyone would like to lower the manual differential rate dramatically or even eliminate the manual differential with better instrumentation and technology. But it seems to hang on, particularly in the context of oncology and pediatrics. Can you give us a progress report on that from your perspective?
Susan Behnke, MT(ASCP), MBA, senior marketing manager, Horiba Medical: A large part of Horiba’s market is in physician office labs, clinics, and secondary reference labs, so it’s exciting to see the same digital morphology and CBC analyzers for high-end automation coming to the lower-volume labs that may be doing only 10 to 15 CBCs a day. In the past 20 years, hematology analyzers have changed in their ability to report out new parameters as well as new flags to help guide technologists on what they see, if they go to the scope or use digital morphology for confirmation.
Kevin Croghan, what is your reaction to what you’ve heard, in particular about the need to have instrument and other solutions that scale within large networks and systems?
Kevin Croghan, MBA, director, hematology strategy and solutions, Sysmex America: We continue to hear concerns from our customers about staffing. We have full confidence in the pathologists and customers we work with, and our goal is to continue to help. We’ve asked folks, do you want more information? Do you want more parameters? Going forward, it’s about how we can help people understand the information we’re providing and help them make quicker decisions. Maybe a guidance rating for results can be provided in the future. It’s also about our main partner, CellaVision, with its digitalized feathered edge.
Dr. Vergara-Lluri, we hear from industry about its great efforts in automation, artificial intelligence, data handling, et cetera. I assume these are important to you, but can you also talk about how staffing shortages are never out of mind, given tight budgets and the absence of personnel?
Ria Vergara-Lluri, MD, associate professor of clinical pathology, Keck School of Medicine, University of Southern California; attending hematopathologist, Los Angeles General Medical Center; and medical director of hematology section, Los Angeles General Core Laboratory: It is first and foremost in our minds, as I’m a laboratory section director for hematology and also a clinical subject matter expert for leading the Department of Health Services in Los Angeles County in standardizing and integrating laboratory hematology for all the facilities in our county’s 10-million-strong Angeleno population. It’s a daunting task to try to take care of all our patients with dropping staffing numbers and expertise.
We’re moving now from an old system to a new one, and there are bumps along the way. As we’re switching to the new platform, I’m trying to figure out ways to streamline the workflow so our technologists are not incredibly taxed as they go about their days. How can we decrease the points at which they have to touch a specimen, whether that’s doing a manual differential or morphology or even figuring out hemolysis, lipemia—what do I do with this? Do I put it in a HemoCue? Do I do a spun hematocrit or platelet clumps?
In listening to our colleagues in the industry, I’m excited about the integration of AI because I think we pre-sent a glut of information—parameters, et cetera—to our customers and the question is, what does it all mean? What does it mean to say there’s a little bit of elliptocytosis or prekeratocytes? We could help clinicians with the integration of the information and the interpretation that comes with it. Right now in hematology we look at discrete parameters, but it would be cool if we could integrate that and say, These are the most likely possibilities. It would be neat if AI could help us and say, Based on this aggregation of information, you’re more likely looking at a hemoglobinopathy versus an iron deficiency anemia.
Can you speak on reference ranges in hematology analysis, because it’s going to be a key to interpretation, autoverification, and the presentation of an AI-enabled result to clinicians?
Dr. Vergara-Lluri (USC): It is on my to-do list because we have such a diverse population in Los Angeles County, not just diversity of ethnicities but also a transgender population for which reference ranges are needed. That’s a gap, and we would appreciate help with ranges from our vendor colleagues. In turn, as customers, our job would be to verify that, because it would be a huge undertaking.
Dr. Pozdnyakova, what’s top of mind for you?
Olga Pozdnyakova, MD, PhD, director, Division of Hematopathology, and professor, pathology and laboratory medicine, Hospital of the University of Pennsylvania: Reference ranges are an important topic and have to be addressed on a national level. It is a daunting task to do it lab by lab. WHO-published reference ranges go back to before we had the information we have now. I led an effort when I was at Brigham with people who had Duffy null phenotype. We know they have an absolute neutrophil count that is lower than that of people who are not Duffy null phenotype. If you’re not aware of that—and I’m sure many labs are not and it depends on your patient population—you subject many patients to unnecessary testing because old reference ranges include everyone. We didn’t have the information; we didn’t know better. So many patients are being worked up for having chronic neutropenia, for example.
I just formed a committee for the standardization of the manual differential review. It’s a big effort of the CAP, International Society for Laboratory Hematology, Society for Hematopathology, and International Council for Standardization in Haematology. The criteria are antiquated and no one has looked at them in years. We know a lot more now. We measure many more parameters. We have better hematology analyzers. We have digital morphology. We know now that some things are clinically significant but some are not. And the time has come to reassess the criteria for reviewing a peripheral blood smear.
We have been using automation and AI to some extent, but not to their full advantage. Can AI or automation be used to put a report together that includes not just the morphology but also some of the parameters that lead the clinician on the right path to work up a patient for iron deficiency anemia or for elliptocytosis or hemoglobinopathy? We need a systematic approach to create a path for how to do that.
Are industry partners already involved in your efforts to standardize the manual differential review?
Dr. Pozdnyakova (Penn): Not yet. We just formed the committee and haven’t met yet.
What comes to mind as we talk about reference ranges and standardization is that the vendors in hematology sit on an enormous amount of data coming in from all the machines installed worldwide. Fernando Chaves, do you have an interest in helping customers around the globe integrate and standardize information that is already in-house in one way or another?
Dr. Chaves (Siemens Healthineers): Yes. We need to do more with the data that exists. Customers are not asking for new parameters anymore; we want more answers from the parameters that exist. That principle encompasses a lot of things. Now that AI technology exists, be it algorithms based on parameters or image analysis algorithms based on digital images, there is an incredible opportunity to help customers more, either with reference ranges that are specific to certain populations or clinical diagnosis and alerts for specific diseases that can be generated from an algorithm that looks at multiple parameters simultaneously or from image analysis. The parameters we report in a CBC in isolation are difficult for a clinician to make sense of, and the information probably hasn’t been used to help patients. There is an opportunity to do that now and we’re excited about it.
We’re also looking into integrating hematology into other image-based solutions in medicine. In the same way that hematology generates images, images are generated by ultrasound, surgical pathology, medical resonance, et cetera, and solutions are coming that look at multiple pieces of information from different disciplines, even outside of the laboratory, outside of hematology. As hematology becomes digital and artificial intelligence starts playing a role in that, we’re going to see our discipline play a much bigger role in the full suite of information that is taken into consideration to give better answers to patients and clinicians.
Everyone’s touching on an important point: We have staff and expertise and pathologist shortages in labs, but the same has to be true for the clinicians who depend on these often complicated results. Sarah Borah, can you speak to that, and do you have thoughts on improving the reporting so it’s timely and comprehensible?
Sarah Borah (Beckman Coulter): I agree on the reference ranges being a partnership with the vendors because we have access to more patient population data.
If you think about all the information a physician receives from laboratory and other tests, especially in hematology with a CBC run on almost everyone who presents to the emergency department, being able to put that data into a clinical decision support system, build an algorithm on it, is something Beckman Coulter is highly invested in, specifically in areas around sepsis and cardiac care. Taking parameters like MDW [monocyte distribution width], incorporating the SIRS [systemic inflammatory response syndrome] score from the patient and patient history and data, and putting it together into a risk stratification score for critical disease states is high on our radar.
Dr. Pozdnyakova (Penn): I work with Beckman, and we are looking into MDW and sepsis. It’s not only important for clinical decision-making, but now there’s a critical shortage of blood culture bottles. The CBC is the most commonly ordered test. Not only do we help make diagnoses, help create a path for the patient and triage what needs to be done next, but we can also help alleviate or overcome supply situations where we don’t have enough reagents because we looked at those cells or have something within the hematology analyzer. I suspect that the next generation of hematology analyzers will have a way to measure cell properties that can tell us what type of cell it is, whether it’s a B lymphocyte or T lymphocyte, for example. That will also help eliminate unnecessary testing, like not performing flow cytometry on every person with lymphocytosis because we measured or assessed the cells with the hematology analyzer and the AI decided the cells are not malignant, they’re not B cells, so we can stop. It’s powerful.
Sarah Borah (Beckman Coulter): I agree with Dr. Pozdnyakova—a shortage of blood culture bottles could have a large, negative impact on patient care. With hematology biomarkers like MDW, they’re measuring a host response to infection. If that value is negative it’s a great indication that a culture is not needed. So we can help conserve blood culture bottles for the patients for whom it’s most appropriate. That’s a hematology parameter that’s providing diagnostic stewardship and guidance for use of blood cultures based on measuring the response to severe infections.
Susan Behnke, are clinicians in clinics and physician offices primed to have a simpler, more comprehensible reporting structure?
Susan Behnke (Horiba): It’s becoming more and more of a challenge especially in rural areas. In critical access hospitals and big clinics in more rural areas that may not have the expertise of a major medical center, the equipment is what’s needed to help drive that information. The combination of some of the parameters can guide and direct probable and possible causes of changes in values.
The physicians who see these patients either at their site or in the clinic are being monitored on surveys that ask, When the patient was at the clinic, did the physician discuss the plan of treatment at time of service?
We need automation for sites that have large volumes in order to identify the samples that are in need of additional review and the normal samples that need to be autoverified. That same approach has now been driven into the lower-volume physician offices and clinics.
Dr. Vergara-Lluri (USC): The staffing shortages are not a problem just for our clinical laboratory scientists but also in the hematopathology sphere. At ADLM I was seeing a drive to shore up the gap in staffing by having PhD clinical chemist colleagues help in the core laboratory. That space is not necessarily within their expertise but they’re being asked to carry this awesome burden. That’s compounded by the fact that many of our clinical hematology colleagues are geared toward oncology and solid cancer therapy. So the population of the benign hematologists—the ones who understand what it means to look at a blood smear and say, “This is much more consistent with a DIC picture than some other etiology”—is also shrinking. That’s where the standardization Dr. Pozdnyakova is talking about and bringing all the technology to bear is going to be helpful for our patients, because the expertise from person to person will be variable. The hope for and excitement about AI is that we’ll put everyone on the same level.
Dr. Pozdnyakova (Penn): There is a disconnect with our clinical colleagues. I don’t think they understand how the technology has evolved and that the tests we provide are accurate and can guide their decisions. They use our laboratory information all the time, but there is still some mistrust about what I’m telling them. A prominent hematologist-oncologist once told me that every CBC should have a manual differential. We know what this field can offer, but we need to educate our colleagues and tell them it’s important to order that marker because it can tell whether your patient has or does not have sepsis.
I’ll give an example with immature platelet fraction on Sysmex. Very few people use it because our clinical colleagues don’t know to order it and don’t know what it offers. We’re developing great tests with digitalization, but if no one uses them, if our clinical colleagues are skeptical about what we’re doing, the tests aren’t helping anyone. So in addition to what we do, we need to make a major effort to educate hematologists or hematologist-oncologists about the results.
Dr. Vergara-Lluri (USC): The onus is on us as laboratory stewards for utilization and for the good of the patient that we communicate with each other. And we don’t. That disconnect is in sharp contrast to the oncologist’s comment that everything has to have a manual smear. The flip side to that is the band count. We have so much data in the literature in pathology and laboratory medicine, but in the clinical sphere they’re still hanging on to it. We could make major headway in communicating with our colleagues. Reticulated hemoglobins, the IPF [immature platelet fraction], the MDW, et cetera—we’re working on many things on the laboratory side that we need to let our clinical colleagues know about.
Dr. Pozdnyakova, do hematologist-oncologists in their residency or fellowship training spend a number of weeks with you?
Dr. Pozdnyakova (Penn): Unfortunately they do not spend a lot of time with us. I’m sure they’re busy because the field of heme-onc has exploded with new drugs and new trials.
Dr. Chaves (Siemens Healthineers): When we start giving clinicians AI-based clinical answers rather than what is today a data dump of parameters that they don’t know what to do with, that alone will be a major opportunity to alleviate the disconnect between clinicians and the information from the laboratory. Papers are starting to come out on how to pull all the data together and give answers. It’s fascinating. I recently saw a paper on Scopio images and they were using morphology of lymphocytes to predict remission in patients treated for recurrent diffuse large B-cell lymphomas, for example—a random clinical question that can be answered based on cell morphology [Fridberg G, et al. Cancers (Basel). 2023;15(23):5611].
Once we are in a position where we can give clinicians real answers to clinical questions and not give them data and expect them to put it together, much of this challenge will be alleviated organically, and that’s exciting.
We’re getting close to a laboratory role that Michael Laposata [professor and chair, Department of Pathology, University of Texas Medical Branch at Galveston] has articulated well, which is instead of throwing data over the wall at the clinician, let’s give them a comprehensive report and recommendations. Kevin Croghan, do you see that beginning to emerge in hematopathology?
Kevin Croghan (Sysmex): I hope so. I think all of us here would like to have more access to clinicians and pathologists to be able to work together on what the future solutions are. We need to feel comfortable in the industry about being able to provide direction based off some of the results. We’re not providing a yes, this; no, that. We’re providing a general direction on something you may need to investigate further.
Susan Behnke (Horiba): We just came back from ADLM, which was all lab and had all the new technologies coming out. But there’s also ASCO and ASH, and laboratory doesn’t attend those meetings. So there’s a disconnect, like everyone has said, between the laboratory and the clinicians and how and what to do with the data and information they’re receiving.
Dr. Pozdnyakova (Penn): It’s also the papers. There is pathology literature and there is hematology-oncology literature. If you’re lucky, if your project or idea is good enough or close enough to hematology, you’ll publish a blog and maybe it will be noticed by clinical colleagues. Other than that, they don’t read our journals and we would love to but don’t have time to read theirs.
Jim Crawford [former pathology chair at Northwell] published an academic study of this dilemma—clinicians do not avail themselves of a lot of pathology knowledge. Dr. Vergara-Lluri, what’s a typical consult you might have with a hematologist-oncologist in your system?
Dr. Vergara-Lluri (USC): We reach out to our physician colleagues a lot. We’ll tell them this is APL [acute promyelocytic leukemia] before they’ve even heard of the patient—that’s how fast we jump on things. When they come to us, it’s on benign things. They’ll say, “This is a patient who has a fever of unknown origin. I don’t know what’s going on, and we’d like you to figure it out.” With the diminishing expertise on the clinician side, the hematologists should be the experts on the smears. They are experts in a way that they can integrate the clinical information easily alongside all the labs and see patients, et cetera. But every once in a while, they’ll come to me and say, Do you know what this is? I’ll say, for example, there’s an unusual cell here, so let’s get a bone marrow.
I’m interested in getting a standardized interpretation that doesn’t necessitate intervention by an expert hematopathologist rather than data dumping. We’re providing, “This is the differential diagnosis—look into this,” rather than just “anisocytosis 1+, stomatocytes 2+.” Let’s give meaningful information.
Kevin Croghan (Sysmex): Or just be able to highlight that you don’t have to look at this patient anymore.
Yes. To rule out suspicion is an estimable clinical move if you can do it with great confidence in science.
Dr. Chaves (Siemens Healthineers): With digitalization, we are going to tremendously improve the analytical performance of some of the measurements. You can’t get much information or clinical insight out of schistocytosis when you measure it only as 1+, 2+, 3+. With digital solutions we can have a real schistocyte count. No one goes immediately from 0 schistocytes to even 1+ schistocytosis; at that point 1+ already means thousands of schistocytes per million red cells. The same way high-sensitivity troponin opened the door for many new clinical uses for the test, we can, with digital solutions, count many of these morphologic features with better precision and better sensitivity. It’s not difficult to envision the clinical uses that will be uncovered through this improved analytical performance that can exist only through digital hematology.
Would anyone like to raise an important point or question that we haven’t discussed?
Sarah Borah (Beckman Coulter): I’m interested in hearing from the experts about the potential impact of AI and digital morphology now with bone marrow samples, as I can imagine the high level of expertise needed to read a bone marrow. We’re starting to see the AI, similar to anatomic pathology, flow into clinical pathology with the ability to look at tissues like bone marrow. Is it helpful?
Dr. Pozdnyakova (Penn): It’s going to be very helpful, no doubt. I cannot wait to implement it in my lab.
Dr. Vergara-Lluri (USC): I view myself as a community pathologist serving county patients, and I worry about resource-limited settings and the cost. Bringing costs down eventually, as there’s more competition, would be important.
Susan Behnke (Horiba): We haven’t talked today about the studies the FDA requires for new analyzers and what you are able to state about the data generated. The FDA hired many analysts who are less strong clinically. To move information from “This is data” to “This is what it means,” we’ll need to work more closely with the FDA to get that piece of information through as well.